Bug 135263

Summary:	[testcase] Extraneous harmful encoding of Javascript-escaped Unicode strings
Product:	[Applications] konqueror	Reporter:	Rémi Denis-Courmont <remi>
Component:	khtml ecma	Assignee:	Konqueror Developers <konq-bugs>
Status:	VERIFIED WORKSFORME
Severity:	normal	CC:	finex, maksim
Priority:	NOR
Version:	unspecified
Target Milestone:	---
Platform:	Debian testing
OS:	Linux
Latest Commit:		Version Fixed In:
Sentry Crash Report:

Description Rémi Denis-Courmont 2006-10-07 21:14:31 UTC

Version:           3.5.5 (using KDE KDE 3.5.5)
Installed from:    Debian testing/unstable Packages
Compiler:          GCC 4.1 Debian 
OS:                Linux

When faced with characters outside the scope of ISO-8859-1, Javascript function escape will properly use the %uXXXX format (where XXXX is the hexadecimal notation for the Unicode code point). This is correct.

Unfortunately, when Konqueror actually tries to access an URL with such an encoded string as a parameter, it will replace %uXXXX with %25uXXXX. Therefore, the server will merely decode '%25' to '%' and will leave 'uXXXX' as is. This is incorrect.

Konqueror must not escape '%uXXXX' characters sequences manually when these are found in a requested URL. Firefox has a correct behavior here.

This is also Debian bug #359784.

As an example: http://www.remlab.net/tmp/

In Mozilla, clicking "Click here" yields the original message back. Konqueror, mangles it.


Solution: in whatever function decides whether or not some part of the URL shall be URL-encoded or not, please do conside '%uXXXX' for encoding, and pass it unchanged instead.

Comment 1 FiNeX 2008-05-20 20:20:01 UTC

Confirmed on 3.5.9 and KDE 4 (trunk)

Comment 2 Maksim Orlovich 2009-02-27 18:11:59 UTC

Looks fine these days..

Comment 3 Rémi Denis-Courmont 2009-02-28 15:43:26 UTC

I still have the problem with Konqueror from KDE 3.5.9. I'll check 4.2 on monday.

Comment 4 Rémi Denis-Courmont 2009-03-05 16:46:27 UTC

Yeah, it works with Konqueror 4.2